Perch 2.0 transfers 'whale' to underwater tasks
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908690331729920 |
|---|---|
| author | Burns, Andrea Harrell, Lauren van Merriënboer, Bart Dumoulin, Vincent Hamer, Jenny Denton, Tom |
| author_facet | Burns, Andrea Harrell, Lauren van Merriënboer, Bart Dumoulin, Vincent Hamer, Jenny Denton, Tom |
| contents | Perch 2.0 is a supervised bioacoustics foundation model pretrained on 14,597 species, including birds, mammals, amphibians, and insects, and has state-of-the-art performance on multiple benchmarks. Given that Perch 2.0 includes almost no marine mammal audio or classes in the training data, we evaluate Perch 2.0 performance on marine mammal and underwater audio tasks through few-shot transfer learning. We perform linear probing with the embeddings generated from this foundation model and compare performance to other pretrained bioacoustics models. In particular, we compare Perch 2.0 with previous multispecies whale, Perch 1.0, SurfPerch, AVES-bio, BirdAVES, and Birdnet V2.3 models, which have open-source tools for transfer-learning and agile modeling. We show that the embeddings from the Perch 2.0 model have consistently high performance for few-shot transfer learning, generally outperforming alternative embedding models on the majority of tasks, and thus is recommended when developing new linear classifiers for marine mammal classification with few labeled examples. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_03219 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Perch 2.0 transfers 'whale' to underwater tasks Burns, Andrea Harrell, Lauren van Merriënboer, Bart Dumoulin, Vincent Hamer, Jenny Denton, Tom Machine Learning 68T07 I.2.1 Perch 2.0 is a supervised bioacoustics foundation model pretrained on 14,597 species, including birds, mammals, amphibians, and insects, and has state-of-the-art performance on multiple benchmarks. Given that Perch 2.0 includes almost no marine mammal audio or classes in the training data, we evaluate Perch 2.0 performance on marine mammal and underwater audio tasks through few-shot transfer learning. We perform linear probing with the embeddings generated from this foundation model and compare performance to other pretrained bioacoustics models. In particular, we compare Perch 2.0 with previous multispecies whale, Perch 1.0, SurfPerch, AVES-bio, BirdAVES, and Birdnet V2.3 models, which have open-source tools for transfer-learning and agile modeling. We show that the embeddings from the Perch 2.0 model have consistently high performance for few-shot transfer learning, generally outperforming alternative embedding models on the majority of tasks, and thus is recommended when developing new linear classifiers for marine mammal classification with few labeled examples. |
| title | Perch 2.0 transfers 'whale' to underwater tasks |
| topic | Machine Learning 68T07 I.2.1 |
| url | https://arxiv.org/abs/2512.03219 |